After much discussion in the latter part of 2025, the AI stock bubble has already burst. John Higgins, chief markets economist at Capital Economics, comes to that conclusion. What’s still brewing is what worries him more.
A bubble occurs when assets are valued much beyond their fundamental value, as shown when share prices skyrocket despite solid proof of great financial results. If you judge whether or not a bubble exists based on how stretched or otherwise its valuation is, there is a case to be made that the bubble has popped, Higgins explained.
Higgins discovered that the ratio of current share price to earnings per share has increased over the past three years for information technology and the rest of Big Tech, indicating inflated values, in a note to clients released this week. However, that price-earnings ratio started to decline about October 2025, and it is currently the lowest since the pandemic. The price-earnings ratio was far higher during the dotcom bubble around the turn of the century, surpassing 150% for the IT sector in the early 2000s and peaking at about 75% in late 2024, according to Higgins.
AI valuations have actually risen. According to statistics from tech market intelligence platform CB Insights, there were 498 AI unicorns with a total worth of $2.7 trillion as of fall 2025, with 100 of them created in 2023 or later. More than 1,300 AI businesses have a valuation greater than $100 million. According to CFO Sarah Friar, OpenAI’s valuation rose to $730 billion last month from $500 billion in October, less than six months ago.
But the tech industry has returned to reality, partly due to the “SaaSpocalypse,” a sharp decline in software-as-a-service (SaaS) stocks as investors worry that agentic AI would quickly supplant conventional software business models. Since the start of the year, both Salesforce and ServiceNow have lost around 30% of their respective values.
“That software services industry group was one of those sectors that was relatively vulnerable to that rollout of AI,” Higgins explained, adding that investors had sort of focused on it. “As a result, we saw a significant decline in the valuation of that particular sector.”
Higgins contended that the SaaS sector is not the only one suffering. Additionally, the semiconductor industry has recently slowed down due to a shortage of chips caused by strong demand and supply chain issues brought on by recent geopolitical tensions like tariffs and the conflict in Iran.
The next AI bubble will be rare
According to Higgins, the tale of these businesses’ challenges might conceal another bubble. In recent years, the profitability of tech companies have skyrocketed, prompting concerns about how sustainable this level of development may be. According to Bloomberg Intelligence, the Magnificent Seven will have earnings growth of about 18%, while the other 493 S&P 500 businesses will see growth of 11%. Nvidia reported $68.1 billion in revenue for its fourth quarter last month, up 73% from the previous year.
Higgins stated, “There might be one [bubble] actually in the fundamental side of things, which is quite rare.” “A bubble is typically thought of as something where the price has become out of line with the fundamentals itself. In this instance, the earnings themselves can be the source of the bubble. Higgins was alluding to the primary justification used by proponents of the anti-bubble movement, which is the massive profits generated by the largest public tech companies that control the Magnificent Seven. To put it another way, he wants to know what happens if these profits decline.
AI earnings may soon hit a cliff and cause a market drop for a few reasons. First, according to Higgins, demand for AI might be far lower than first projected, so tech companies will have to deal with Goldman Sachs’ projected $539 billion in AI capital expenditures for 2026. According to McKinsey, 88% of businesses report using AI on a daily basis, but adoption may be stagnating because workers are worried about losing their jobs to the technology.
Higgins indicated that the longer the economy remains in a vulnerable state, the higher the risk to AI revenues. The ongoing Iran war has interrupted helium production in Qatar, which accounts for roughly one-third of the world’s supply of the odorless gas essential to make computer chips. Not only have data centers been targets of attacks during the conflict, but energy prices may also increase the input costs of these facilities.
According to Higgins, even if the demand for AI isn’t actually declining substantially, a downturn in the economy as a whole could have an impact on the stock market and the earnings of companies that are profiting from the implementation of AI.
